For years, machines have been used to scan parcels as they travel along a conveyor. Over-the-belt optical character recognition (OCR) systems have been recently developed that can read indicia, such as a typed or hand-written destination address on parcels to be shipped. Parcel delivery companies, such as United Parcel Service, ship millions of parcels every day. These parcel delivery companies make extensive use of OCR systems to read the destination address labels on parcels to facilitate sorting and routing the parcels to their proper destinations.
The fundamental physical components of an OCR system are a scanner and a character recognition system including a central processing unit (CPU), a computer memory, and a sophisticated character recognition program module. The scanner is typically an optical camera, such as a charge-coupled device (CCD) array, that captures an image of the destination address on the parcels as they travel past the scanner on the conveyor. Generally, a continuous video image of the conveyor carrying the parcels is captured by the scanner, which video image is converted into digital format and transmitted to the character recognition system. But only a small part of the video image, such as the portions including the destination addresses of the parcels, needs to be processed by the character recognition system. The OCR system, therefore, must have some way to identify the portions of the video image that need to be processed by the character recognition system.
One approach is to store the entire video image created by the scanner, and later parse out the portions of the video image that need to be processed by the character recognition system. But a continuously running scanner generates an enormous amount of video data. This data is formatted as a continuous bit map of the conveyor as the conveyor carries parcels past the scanner, which bit map inherently convey information about the spatial relationship of the pixels of the image. Storing this continuous bit map requires an enormous amount of computer memory. It is therefore advantageous to reduce the memory storage requirement.
Data compression is one technique for reducing the memory storage requirement. The video data may be compressed for storage using any of a variety of well known data compression methods, such as run length encoding. These data compression techniques, however, alter the bit-map format of the data. This is undesirable because it is advantageous for the character recognition program module to operate on bit maps that allow easy access to information regarding neighborhoods around individual pixels. The compressed data must therefore be uncompressed, typically into a frame buffer, for processing by the character recognition program module. Compressing the video data, for storage, and then uncompressing the video data for processing, burdens the CPU and slows the character recognition process.
Real-time extraction of the desired portions of the video data is another technique for reducing the memory storage requirement. Indeed, real-time data extraction is a very effective technique because most of the video data created by the continuously running scanner is a useless image of the conveyor and the non-indicia bearing areas of the parcels moving along the conveyor; only a small percentage of the data includes the destination addresses of the parcels to be shipped. Therefore, extracting only small portions of the video data, such as relatively small areas covering the destination addresses, greatly reduces the memory storage requirement and speeds up the character recognition process.
Systems have been developed for triggering a video camera system so as to store only desired video images. For example, Tonkin, U.S. Pat. No. 4,742,555, describes a mechanical limit switch, optical sensor, or magnetic sensor that triggers a video system to capture and store an image of a parcel as the parcel reaches a predetermined location along a conveyor. But the system described by Tonkin would have a significant drawback if applied to a parcel shipping system. This is because the system described by Tonkin captures an image of the entire parcel; is not operative for capturing only a specific portion of the image, such as the destination address. In a parcel shipping system, the destination address must be captured for sorting and routing purposes, but other indicia on the parcel, such as the return address, is not needed to route the parcel to its proper destination. It is therefore advantageous to identify the destination address prior to storing the image of the parcel, so that only the portion of the image containing the destination address may be stored in the computer memory.
Several difficulties are encountered, however, in attempting to identify the destination addresses on various parcels traveling on a conveyor. First, the destination addresses may vary in size, and may be in different locations on different parcels. Second, the parcels themselves may vary in size, shape, and position on the conveyor. Thus, the exact position of a destination address on a parcel cannot be determined by simply detecting the edge of the parcel using a limit switch or sensor, as described by Tonkin.
Systems have been developed for storing video images of selected portions of parcels traveling of a conveyor. For example, Kizu et al., U.S. Pat. No. 4,516,265, describes a two camera system that reads the postal (zip) codes on envelopes traveling on an envelope transport system. The system includes a low resolution prescanner that coarsely scans the surface of the envelope. The position of the destination address block is determined from the coarse scan, and the coordinates of the destination address block with respect to the leading edge of the envelope are then passed to a second, high-resolution camera system. The second camera system stores an image of the destination address block by first detecting the leading edge of the envelope. The second camera system begins storing an image of the destination address block when the block reaches the second camera, and stops storing the image when the block moves past the second camera. A postal code reader subsequently processes the high-resolution scan to read the postal code.
Another example is disclosed in the commonly owned U.S. patent application Ser. No. 08/536,512, entitled "Two Camera System for Locating and Storing Indicia on Conveyed Items." This application describes a two camera system that reads the destination addresses on parcels traveling on a conveyor. A fluorescent ink fiduciary mark is superimposed relative to the destination address on a parcel. A first camera captures an image of the fiduciary mark, the position and orientation of which is ascertained. The position and orientation of the fiduciary mark is then used to extract an image of the destination address from a video data signal created by a second camera, which is positioned downstream from the first camera. The image of the destination address is stored in a computer memory for subsequent processing by a character recognition system.
The two camera systems described above are very effective at minimizing the amount of video data that must be stored in an OCR system. They are, however, rather expensive systems that are best suited for very high-speed parcel handling systems. The cost associated with these systems may not be justified for many lower-speed parcel handling systems. There is, therefore, a need for a less expensive system for minimizing the amount of video data that must be stored in an OCR system. In particular, there is a need for an inexpensive indicia reader system that is suited to low- to medium-speed parcel handling systems.